Re: [PATCH 3/6] core: Add stream id list in the cell config

['Pratyush Yadav' via Jailhouse]

On 02/07/19 9:11 PM, Jan Kiszka wrote:
> On 02.07.19 16:36, Pratyush Yadav wrote:
>> From: Nikhil Devshatwar <nikhil...@ti.com>
>>
>> When IOMMU drivers are supported, it will setup the IO address
>> translation tables unique for each DMA context in the system.
> 
> To make it clearer: When ARM IOMMUs are supported...

Will update.

>>
>> A typical DMA context is identified by an integer called stream id.
>> To setup the correct IOMMU mapping, hypervisor should know
>> list of all the streamIDs that should be setup in the IOMMU.
>>
>> Add an array of stream IDs in the cell config, bump up the config revision.
>> Change the python struct unpacking logic in the tools to reflect
>> revised config structure.
>>
>> [p-yad...@ti.com: Add size define for stream_ids]
>>
>> Signed-off-by: Nikhil Devshatwar <nikhil...@ti.com>
>> Signed-off-by: Lokesh Vutla <lokeshvu...@ti.com>
>> Signed-off-by: Pratyush Yadav <p-yad...@ti.com>
>> ---
>>   include/jailhouse/cell-config.h | 6 +++++-
>>   tools/jailhouse-cell-linux      | 4 ++--
>>   tools/jailhouse-hardware-check  | 2 +-
>>   3 files changed, 8 insertions(+), 4 deletions(-)
>>
>> diff --git a/include/jailhouse/cell-config.h 
>> b/include/jailhouse/cell-config.h
>> index 198fef5c..caf2fba8 100644
>> --- a/include/jailhouse/cell-config.h
>> +++ b/include/jailhouse/cell-config.h
>> @@ -50,7 +50,7 @@
>>    * Incremented on any layout or semantic change of system or cell config.
>>    * Also update HEADER_REVISION in tools.
>>    */
>> -#define JAILHOUSE_CONFIG_REVISION    10
>> +#define JAILHOUSE_CONFIG_REVISION    11
>>     #define JAILHOUSE_CELL_NAME_MAXLEN    31
>>   @@ -74,6 +74,9 @@
>>     #define JAILHOUSE_CELL_DESC_SIGNATURE    "JHCELL"
>>   +#define JAILHOUSE_MAX_STREAMIDS                32
> 
> At the risk of having asked this already in the RFC:

I thought you meant you wanted to change the hard value of 32 to a size define. 
My bad, sorry.

> Is 32 really enough in the foreseeable future? Or should we rather have some 
> resizable array, analogously to memory regions or PCI devices? That would 
> also allow to set it to 0 on non-SMMU platforms.

Seems like a good number. We don't use more than 8 in our boards. But changing 
it to a resizeable array should not be a problem either.

Now that I think of it, it is probably a better idea. I'll change it.

>> +#define JAILHOUSE_INVALID_STREAMID            (~0)
>> +
>>   /**
>>    * The jailhouse cell configuration.
>>    *
>> @@ -95,6 +98,7 @@ struct jailhouse_cell_desc {
>>       __u32 pio_bitmap_size;
>>       __u32 num_pci_devices;
>>       __u32 num_pci_caps;
>> +    __u32 stream_ids[JAILHOUSE_MAX_STREAMIDS];
>>         __u32 vpci_irq_base;
>>   diff --git a/tools/jailhouse-cell-linux b/tools/jailhouse-cell-linux
>> index 49babd92..4b6f504b 100755
>> --- a/tools/jailhouse-cell-linux
>> +++ b/tools/jailhouse-cell-linux
>> @@ -564,8 +564,8 @@ class Irqchip:
>>       class Config:
>> -    _HEADER_FORMAT = '6sH32s4xIIIIIIIIIQ8x32x'
>> -    _HEADER_REVISION = 10
>> +    _HEADER_FORMAT = '6sH32s4xIIIIIIII128xIQ8x32x'
>> +    _HEADER_REVISION = 11
>>         def __init__(self, config_file):
>>           self.data = config_file.read()
>> diff --git a/tools/jailhouse-hardware-check b/tools/jailhouse-hardware-check
>> index b86756ac..afd1139b 100755
>> --- a/tools/jailhouse-hardware-check
>> +++ b/tools/jailhouse-hardware-check
>> @@ -136,7 +136,7 @@ class Sysconfig:
>>       X86_MAX_IOMMU_UNITS = 8
>>       X86_IOMMU_SIZE = 20
>>   -    HEADER_REVISION = 10
>> +    HEADER_REVISION = 11
>>       HEADER_FORMAT = '6sH'
>>         def __init__(self, path):
>>
> 
> Jan
> 

-- 
Regards,
Pratyush Yadav

-- 
You received this message because you are subscribed to the Google Groups 
"Jailhouse" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to jailhouse-dev+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/jailhouse-dev/0bfd4d29-4f35-5346-209d-cafcdbdb2151%40ti.com.
For more options, visit https://groups.google.com/d/optout.